TO:UTC

FROM:BiligsaikhanBatjargaletal.(forwardedbyDeborahAnderson,ScriptEncoding
Initiative,UCBerkeley)
RE:MongolianScriptRenderingIssues
DATE:30July2010
Theattachedpaperisadraftandreflectscurrentresearchbytheauthors.NOTE:Itisnottobe
distributedoutsidetheUTC/L2becauseitisintendedforeventualpublication.However,the
authorswelcomecommentsandquestions.Pleaseaddressfeedbackandquestionsto:
biligsaikhan@gmail.com.
Itishopedthatrenderingissuescanberesolvedsoon,particularlyinlightoftherecent
announcementbythePresidentofMongoliathatrequirestheuseoftraditionalMongolian
script(alongsideCyrillic)oncertainofficialdocuments.Alinktotheannouncement(in
Cyrillic)is:http://www.president.mn/mongolian/node/903.
Anabbreviated,unofficialEnglishtranslationisprovidedbelow(from
http://www.montsame.mn/index.php?option=com_news&mt=normal_news&tab=201007&task=
news_detail&ne=264):

A Survey on Rendering Traditional Mongolian Script

Biligsaikhan Batjargal

, Fuminori Kimura

and Akira Maeda

Graduate School of Science and Engineering, Ritsumeikan University

College of Information Science and Engineering, Ritsumeikan University

1-1-1 Noji-Higashi, Kusatsu, Shiga 525-8577, Japan
biligsaikhan@gmail.com, {fkimura, amaeda} @is.ritsumei.ac.jp
Abstract. This paper discusses the rendering issues of complex text layouts,
particularly traditional Mongolian script. Solving the rendering issues of
complex text layouts is the key fundamental challenge to succeed in the future
development of digital libraries. Recently some standards such as Unicode and
OpenType format have been implemented and supported widely. Furthermore,
traditional Mongolian script has been standardized in Unicode. In this paper, we
analyzed existing OpenType fonts and their rendering schemes for traditional
Mongolian script. We found some errors, and discovered grammatical rules,
which are not documented in international standards. None of the existing
OpenType fonts was complete. Lastly, this paper provides some improvements
and recommendations for future development.

Keywords: Traditional Mongolian Script, Unicode, Encoding, OpenType,
Complex Text Rendering, Digital Library
1 Introduction
In the past decade, much has been initiated in digital library research and
development in the Asia-Pacific region, and the importance of digital library systems
that preserve cultural heritage have also increased. Internationalization and digital
cultural heritage preservation in the Asia-Pacific region require digital library systems
to support various ancient or modern complex text scripts. Also writing systems of
Asia-Pacific that requires complex transformations between text input and text
display for proper rendering are needed. For these complex text scripts the way text is
stored is not mapped to the way it is displayed in a straightforward fashion like
western scripts. Examples of such writing systems are the Arabic, traditional
Mongolian and Brahmic (Indic) family such as Devanagari or Dravidian scripts or the
Thai alphabet. Rendering complexities of some scripts is well documented and
surveyed, though traditional Mongolian script is not.
The traditional Mongolian script digital library (TMSDL)
1
[1][2][3] was developed
to preserve over 800 years of old historical records written in traditional Mongolian
script. One of the challenges for the TMSDL is to display and render documents
written in traditional Mongolian script correctly. Once, due to poor support for
traditional Mongolian script at the operating system level, Garmaabazar et al.[3]

1
http://www.dl.is.ritsumei.ac.jp/tmsdl/
2 Biligsaikhan Batjargal, Fuminori Kimura and Akira Maeda
developed a conversion algorithm to display contents in traditional Mongolian script.
However, recently in Windows Vista and in later versions, especially in Windows 7,
support for traditional Mongolian script and the input locale has been added.
Microsoft Windows multilingual text rendering engine Uniscribe (Unicode script
processor) renders OpenType fonts that can handle the diverse behaviors of all the
world's writing systems, including traditional Mongolian script. Thus, this paper
surveys the key fundamental component for digital library systems, the rendering of
traditional Mongolian script, which is considered to be one of the most complex
writing systems in the world.
2 The Traditional Mongolian Script
The traditional Mongolian script is written vertically from top to bottom in columns
advancing from left to right. This script is the writing system for the Mongolian
language and has three derivative scripts - Todo, Manchu, and Sibe (Xibe). The Todo
script is used in Oirat and Kalmyks. The Sibe script is used in Xinjiang, in the
northwest of China. Similar to Arabic, traditional Mongolian is a contextual script
where letters are cursively joined and have initial, medial, and final presentation
forms for the same letter. In most cases, the letters join together along a vertical stem,
but in the case of certain consonants, which lack a trailing vertical stem, they may
form a single ligature with a following vowel. In addition to these cursive and
positional forms, many letters also have variant forms used in accordance with
spelling and grammatical rules. Thus, the traditional Mongolian script is regarded as
complex; encoding complex script features, as well as understanding the layout
features and rules exclusively related to the script, are crucial for researchers and
developers in the digital library community.
2.1 Grammar and Rules of Traditional Mongolian Script
Some important elements and grammatical rules of traditional Mongolian script are
explained below:
2.1.1 Vowel Harmony
Traditional Mongolian script has a characteristic feature of vowel harmony,
whereby a word can only contain either back vowels (a, o, u) or front vowels
(e, oe, ue), but not both at the same time, with the exception only of a certain
limited set of wordsthe majority of which are foreign words. That is to say, the
vowels in a word are either all masculine and neuter (that is, back vowels plus i)
or all feminine and neuter (that is, front vowels plus i). Words that are written
with masculine/neuter vowels are considered to be masculine, and words that are
written with feminine/neuter vowels are considered to be feminine. The vowel i is
considered neutral and can therefore occur in both front and back voweled words, but
when i occurs in all syllables the word is considered to be front voweled and
behaves as feminine (for example, taking feminine suffixes).
A Survey on Rendering Traditional Mongolian Script 3
Vowel harmony is an important element of the encoding model, as the gender of a
word determines the glyph form of the velar series of consonant letters for traditional
Mongolian script. In traditional Mongolian script, the velar letters (qa and ga) have
both masculine and feminine forms. The masculine and feminine forms of these
letters have different pronunciations. When one of the velar consonants precedes a
vowel, it takes the masculine form before masculine vowels, and the feminine form
before feminine or neuter vowels. In the latter case, a ligature of the consonant and
vowel is required. When one of these consonants precedes another, or is the final
letter in a word, it may take either a masculine or feminine glyph form, depending on
its context. Consequently, the rendering system should automatically select the
correct gender form for these letters based on the gender of the word [4][5][6]. Vowel
harmony in traditional Mongolian script is illustrated in Fig. 1.

Fig. 1. Traditional Mongolian Script Gender Forms [4].
2.1.2 The a and e in a Word-final Position
In traditional Mongolian script, the letters a and e in a word-final position may
take a forward tail or backward tail form depending on the preceding consonant
that they are attached to. In some words, a final letter a or e is separated from the
preceding consonant by a narrow gap; in that case the vowel always takes the
forward tail form, and the a or e is an integral part of the word stem. Whether a
final letter a or e is joined or separated is purely lexical and is not a question of
varying orthography.
2.1.3 Syllable Closed Consonants
The traditional Mongolian script has another important rule syllable closed
consonants. Consonants which exist at the end of a syllable or a word, and the
following letter is not a vowel, are considered to be syllable closed consonants. This
rule could be interpreted as a consonant final syllable rule.
Depending on whether a consonant is syllable closed or not, the glyph form of that
consonant varies, taking either a masculine, feminine or variant glyph form. Also,
there are two types of syllable closed consonants: soft and hard. Hard syllable
closed consonants are the traditional Mongolian letters ba, qa, ga, ra, sa, ta,
and da. Letters na, la, ma, wa, and ang are soft syllable closed consonants.
Syllable closed consonants affect the form of the following suffixes [6]. In this way,
the rendering system should automatically select the correct glyph form for syllable
closed consonants.
4 Biligsaikhan Batjargal, Fuminori Kimura and Akira Maeda
2.1.4 Case Suffix
In traditional Mongolian script, case suffixes are separated from the stem of a word or
from other suffixes by a narrow gap. Suffixes have masculine and feminine pairs (for
example, dur/-tur and dr/-tr), and a stem may take receive suffixes.
Any attached suffixes are considered to be an integral part of the word as a whole.
A suffix affects the form of the preceding letters. What`s more, the final letter of the
stem or suffix preceding the particular suffix takes the final positional form, whereas
the first letter of the particular suffix usually takes a medial form or a final form
(single letter suffixes), depending on the particular suffix.
Some special cases where the first letter of the suffix takes a normal initial form
and a variant initial form are listed in Table 1.
Table 1. Some special cases suffixes of traditional Mongolian [5][6].
Case
Case
suffixes
Shape of the
Initial glyph
Special attention
Dative-
Locative
Case

tur/tr the initial form
of ta
Added to words ending in vowels
and soft syllable closed
consonants tu/t

dur/dr the initial form
of da
Added to words ending in hard
syllable closed consonants
du/d
Ablative
Case

aa/ee
the initial form
of e
Pronunciation and encoding varies
according to vowel harmony
Comitative
Case

lua
the initial form
of la
Added to masculine words

lge Added to feminine words

tai/tei
the initial form
of da
Pronunciation and encoding varies
according to vowel harmony

Any rendering system should consider all the above rules and also must select the
correct glyph form for a letter according to the grammatical rules of traditional
Mongolian.
2.2 Traditional Mongolian Script in the Unicode Standard
From the Unicode Standard version 3.1 to the latest Unicode Standard version 5.2.0,
traditional Mongolian script has been standardized in Unicode and isolated form for
the vowels, and the initial form for the consonants are encoded at the range of
U+1800-U+18AF [4]. Other encoding standards were surveyed by Garmaabazar et al.
[3]. Traditional Mongolian script has also been standardized in the Chinese standard
GB 18030. However, some implementations that follow the GB 18030 standard, such
as Menksoft Mongolian IME
2
use the Private Use Areas (PUA) at the range of
U+E234-U+E34F, instead of using basic characters (U+1800-U+18AF) of Unicode
for storing the traditional Mongolia text.

2
http://en.wikipedia.org/wiki/Menksoft_Mongolian_IME
A Survey on Rendering Traditional Mongolian Script 5
2.2.1 Representative Glyphs
The encoded characters in the Unicode range at U+1800-U+18AF are the isolated
forms for the vowels and the initial forms for the consonants. Letters that share the
same glyph forms are distinguished by using different positional forms for the
Mongolian code range. For example, the representative glyph for U+1823 (Mongolian
letter o) is in the isolated form, whereas the representative glyph for U+1824
(Mongolian letter) u is in the initial form. The various positional and variant glyph
forms of a letter are considered as presentation forms. It is the responsibility of the
rendering system to select the correct glyph form for a letter according to its context.
Thus, having a robust rendering algorithm is vital for displaying traditional
Mongolian script correctly.
2.2.2 Variant forms: Free Variation Selectors
Free variation selectors are encoded in Unicode for traditional Mongolian script,
when a glyph form cannot be predicted algorithmically by the rendering system.
Those are:
U+180B Mongolian free variation selector one (FVS1);
U+180C Mongolian free variation selector two (FVS2); and
U+180D Mongolian free variation selector three (FVS3).
The user needs to append an appropriate variation selector to the letter to indicate
to the rendering system which glyph form is required. These format characters
normally have no visual appearance. When required, a free variation selector
immediately follows the base character it modifies. This combination of base
character and variation selector is known as a standardized variant
3
[4].
2.2.3 Narrow No-Break Space
The narrow no-break space (NNBSP)U+202F is encoded to define traditional
Mongolian suffixes as an integral part of the word as a whole. Basically, a line break
opportunity does not occur before a suffix, and whitespace is represented when using
NNBSP [4].
2.2.4 Mongolian Vowel Separator
The Mongolian vowel separator (MVS)U+180E is used to represent the whitespace
that separates a final letter a or e from the rest of the word. MVS is very similar in
function to NNBSP, as it divides a word with a narrow non-breaking whitespace.
Whereas NNBSP marks off a grammatical suffix, the a or e following MVS is not
a suffix but an integral part of the word stem. For example, the word qana without a
gap before the final letter a means, the outer casing of a vein, whereas the word
qana with a gap (MVS) before the final letter a means, the wall of a tent, as
shown in Fig 2. The words qana are encoded U+182C, U+1820, U+1828, U+1820
and U+182C, U+1820, U+1828, U+180E, U+1820 respectively.

3
http://www.unicode.org/Public/UNIDATA/StandardizedVariants.html
6 Biligsaikhan Batjargal, Fuminori Kimura and Akira Maeda
The MVS always selects the forward tail form of a following vowel a or e.
Also, it may affect the form of the preceding letter. The particular form that is taken
by a letter preceding an MVS depends on the particular letter and in some cases on
whether traditional or modern orthography is being used [4].

Fig. 2. Mongolian Vowel Separator [4].
2.3 OpenType Format
The OpenType format is a cross-platform compatible font format developed jointly by
Adobe and Microsoft. The OpenType supports widely expanded multilingual
character sets and layout features, which provides richer linguistic support and
advanced typographic control, such as ligatures, glyph substitution, swash variants,
kerning, and more. OpenType fonts allow embedding the traditional Mongolian script
rules in a single file. Microsoft developed the guidelines for creating and supporting
OpenType fonts for traditional Mongolian Script [7].
Recently, several OpenType fonts Code2000, Simsun-18030, Daicing fonts,
Manchu Font 2005, Mongolian Baiti, MongolUsug and MongolianScript were
developed for traditional Mongolian script. We surveyed possible candidate fonts
Mongolian Baiti, MongolUsug, and MongolianScript for testing to render traditional
Mongolian script correctly.
2.3.1 Mongolian Baiti
This font was developed by the Founder Corporation, Peking University. It is
important to note that Microsoft distributed Mongolian Baiti with MS Windows Vista
and Windows 7. The latest version is 5.01. Isolated forms for the vowels and the
initial form for the consonants are encoded at the range of U+1800U+18AF. The rest
of the variant forms of glyphs and presentation forms are stored within the font file
and indexed by Glyph Index (GID).
2.3.2 MongolianScript
This font was developed by Erdenechimeg Myatav, the task force leader of the
proposal Traditional Mongolian Script in the ISO/IEC 10646 and Unicode
Standards
4
. The latest version is 2.0. Variant forms of glyphs or presentation forms
are encoded at the range of U+F300U+F3B0. The ligature set is encoded at the range
of U+F400U+F4C1.

4
http://www.iist.unu.edu/newrh/III/1/docs/techreports/report170a.tgz
A Survey on Rendering Traditional Mongolian Script 7
2.3.3 MongolUsug
This font was distributed with traditional Mongolian script editorVertNote
5
. The
latest version is 2.37. Variant forms of glyphs, presentation forms, and ligatures are
encoded in the PUA at the range of U+E000U+E811.
2.4 Rendering System
Recently, rendering traditional Mongolian script via Unicode and OpenType font has
begun to be supported in most Windows applications (including Microsoft Office
Publisher, most Adobe applications, and Microsoft Office, from version 2003),
Windows Vista, Windows 7 and many Mac OS X applications, including Apple's
own, such as TextEdit, Pages and Keynote. In older versions of Windows, OpenType
support for traditional Mongolian script can be added by updating the Uniscribe
driver to the latest version. In Unix-like systems, there are active developments such
as Pango
6
. The XenoType Technologies is working to release a Mongolian Language
Kit
7
for Mac OS X.
However, no attempt has been carried out to check the rendering systems of
traditional Mongolian script and shaping algorithms of the OpenType fonts.
3 Experiments
We conducted a preliminary experiment to survey the rendering algorithm of
OpenType fonts as well as to check that traditional Mongolian script contents are
displayed correctly. Rendering algorithms of the three Mongolian OpenType fonts:
Mongolian Baiti, MongolianScript, and MongolUsug, are reviewed. In addition to
checking the basic rules of traditional Mongolian script, we tested the following
complex grammatical rules:
Vowel harmony, feminine and masculine words;
Syllable closed consonants;
Case suffixes; and
Usage of free variant selectors.
We rendered text and HTML files with the same words of traditional Mongolian
script in the different OpenType fonts, and compared the rendered results with the
correct forms. We selected over a hundred feminine words with syllable closed
consonants or suffixes. We tested all 43 standardized variants of traditional
Mongolian script. The experimental setup is for the English version of Windows 7,
build 7600.16385 with Internet Explorer version 8.0 and Uniscribe Unicode script
processor version 1.626.

5
http://www.uukhai.com/archives/1152
6
http://www.pango.org/
7
http://www.xenotypetech.com/osxMongolian.html
8 Biligsaikhan Batjargal, Fuminori Kimura and Akira Maeda
3.1 Vowel Harmony, Feminine and Masculine Words and Syllable Closed
Consonants
The rendered results of our experiment, which surveys the rules of traditional
Mongolian script (vowel harmony and syllable closed consonants were unusual as
illustrated in Table 2), resulted in irregular for all fonts. For instance, the syllable
closed consonant ga-U+182D of the feminine word has rendered correctly in some
words, but not in all. MongolUsug in particular has failed to render the syllable closed
consonants. Errors which occurred are highlighted and explained in the last column.
3.2 Suffixes
MongolianScript has failed to render almost all suffixes, though Mongolian Baiti and
MongolUsug have failed on comitative case suffixes. The rendered results of the
grammatical suffixes for traditional Mongolian script are illustrated in Table 3.
Table 2. Rendering the traditional Mongolian script in various OTFs.

A Survey on Rendering Traditional Mongolian Script 9
3.3 Usage of Free Variant Selectors
Mongolian Baiti has failed to render 6 variants, MongolianScript 24 and MongolUsug
12 variants of all 43 forms. The rendered results of the selected standardized variant
[4] for traditional Mongolian script are illustrated in Table 4. Errors which occurred
are highlighted and explained briefly in the last column.
Table 3. Rendering the grammatical suffixes of traditional Mongolian script in various OTFs.

4 Conclusion and Recommendation
In this paper, we surveyed the rendering issues of complex text layouts especially in
regard to traditional Mongolian script. We analyzed existing OpenType fonts and
their rendering schemes for traditional Mongolian script. The study produced some
errors, and revealed grammatical rules, which are not documented in international
standards.
OpenType fonts that we surveyed did not include certain rules of traditional
Mongolian script such as vowel harmony, feminine and masculine words, syllable
closed consonants and case suffixes. We realized that some rules such as syllable
closed consonants and the usage of case suffixes, have not been well documented in
Unicode standard.
All fonts failed to display the variant glyphs with free variant selectors correctly,
which were already standardized in Unicode. In general, Mongolian Baiti was better
10 Biligsaikhan Batjargal, Fuminori Kimura and Akira Maeda
than others with few shortcomings on rendering variant forms, suffixes and feminine
words.
In addition, all fonts need some improvements in their rendering algorithms and all
grammatical rules need to be standardized in international use such as in Unicode.
The guidelines for creating and supporting OpenType fonts for traditional Mongolian
Script [7] need to be updated as well.
Finally, we believe this survey will make some contribution to developers of
digital libraries that utilize complex scripts such as traditional Mongolian script.
Table 4. Rendering the variant glyphs of traditional Mongolian script in various OTFs.

References
1. Garmaabazar, K., Maeda, A.: Retrieval Technique with the Modern Mongolian Query on
Traditional Mongolian Text. In: 9th International Conference on Asian Digital Libraries,
LNCS, vol. 4128, pp. 478--481. Springer (2006)
2. Garmaabazar, K., Maeda, A.: Building a Digital Library of Traditional Mongolian Historical
Documents. In: 7th ACM/IEEE-CS Joint Conference on Digital Libraries, p. 483. ACM
(2007)
3. Garmaabazar, K., Maeda, A.: Developing a Traditional Mongolian Script Digital Library.
In: 11th International Conference on Asia-Pacific Digital Libraries: Universal and
Ubiquitous Access to Information, LNCS, vol. 5362, pp. 41--50. Springer (2008)
4. The Unicode Consortium: The Unicode Standard 5.0. Addison-Wesley (2007)
5. Pugh, ERE., Mongolian Grammar Reference, http://www.linguamongolia.com/vhar1.html
6. Choimaa, Sh.: Mongol bicigiin zov bicix drmiin xuraangui (in Mongolian), The Mongolia
Society special papers, Bloomington, (1991)
A Survey on Rendering Traditional Mongolian Script 11
7. Creating and Supporting OpenType Fonts for the Mongolian Script,
http://www.microsoft.com/typography/otfntdev/mongolot/